Retired, Not Released: Auditing Contingency Before the Final Account

There is a familiar moment on a UK refurbishment or fit-out as practical completion approaches. The cost report lands, the contingency line is still substantially intact, and someone in the room observes that the project is "underspent". Within a fortnight the figure has migrated from the risk column to the savings column, and the next capital plan is being built on money that was never surplus in the first place.

Contingency is not a budget line that happens to be unused. It is the price of not knowing, paid in advance, and it is earned back only when the risks it was bought against have genuinely passed. The distinction matters because the two treatments produce different decisions: one improves the record, the other improves the mood.

What the allowance was actually bought for

A single contingency figure on a cost plan usually conceals several different things. A design development allowance covers information not yet complete at the point of pricing. A construction risk allowance covers ground conditions, existing structure, services diversions and the general friction of building inside an occupied estate. A change allowance anticipates that the client will, in fact, change something. An escalation allowance covers the gap between tender and completion.

Each of these retires on a different timetable and against different evidence. A design allowance is largely spent or released once the last package is at construction status. An escalation allowance retires as the packages are let and the exposure narrows to the remaining trade contracts. A construction risk allowance tied to an unexplored riser or an unrecorded drainage run does not retire until that work is complete and no further penetrations are planned.

Lumping them together and reporting a single percentage remaining is convenient for a dashboard and useless for a release decision.

The retirement test

A risk has retired when the window in which it could occur has closed, not when it has failed to occur so far. That distinction is the whole discipline.

For each risk carried in the register, the cost consultant should be able to state the trigger event that ends the exposure and the evidence that the trigger has been met. Asbestos allowances retire when the affected zone is complete and no further intrusive work is scheduled there — not when the last air test came back clear. Utility connection allowances retire when the supply is energised and the adoption or vesting position is agreed, not when the application was submitted. Party wall exposure retires when the awards are satisfied and the neighbouring works are finished. Watertightness exposure retires at roof completion, subject to the defects period.

This is not bureaucracy for its own sake. It is the difference between saying a risk did not materialise and demonstrating that it no longer can.

Three false retirements to watch for

The first is silence. A risk that has not appeared in the monthly report for four months has not necessarily gone; it may simply have stopped being discussed. The register should show the risk, its owner, and the reason for any change in status. Absence of noise is not evidence of closure.

The second is assumed transfer. A risk is sometimes marked as the contractor's and quietly removed from the client's allowance. That is legitimate only where the contract actually allocates it there, and after the amendments, the carve-outs and the design liability provisions have been read rather than summarised. Under a JCT design and build arrangement the allocation may be wide, but it is rarely total, and the exceptions are usually the expensive ones.

The third is deferral dressed as retirement. On a phased handover, a risk affecting floors three to six has not retired because floors one and two are complete. Where occupation is incremental, the exposure moves rather than ends, and the allowance should move with it.

A release process that survives scrutiny

The practical answer is a contingency release schedule, maintained alongside the cost report and reviewed at the same cadence. For each risk it records the original allowance, the retirement trigger, the evidence provided, the date of retirement and the approver. Released sums then move through the change control or budget revision process in the ordinary way, with a decision recorded, rather than being quietly absorbed into a favourable variance.

Release should also be staged. A first tranche at practical completion, covering risks whose window has clearly closed. A second at the end of the twelve-month defects liability period, once the snagging and latent defect exposure has been tested by a full seasonal cycle. A final residual at final certificate, when the final account is agreed and retention is released. Holding a defined tail back is not excessive caution; it is the only way to avoid reopening a closed budget when a defect appears in month eleven.

Where the client is a fund, a local authority or a body subject to external audit, the schedule also serves as the audit trail. It answers the question that arrives six months later: on what basis was this money returned to the capital programme?

Why banking it as savings distorts the next year

Treating released contingency as savings does three kinds of damage. It flatters the current year's outturn against a baseline that was always cautious, which is a reporting artefact rather than a performance gain. It corrupts the benchmark used to size contingency on the next scheme, because the next cost plan will be built on the assumption that the last one was too generous. And it creates an incentive to run the next project with a thinner allowance, which converts a governance success into a future exposure.

The more accurate framing is that the allowance has done its job. It absorbed the risks that materialised, and the remainder has been returned. That is a good outcome in its own right, and it does not need to be relabelled to be reported.

The record is the control

Contingency released without evidence is indistinguishable, in the ledger, from contingency that was never needed. The difference exists only in the documentation. That is why the discipline sits less in the release decision than in the schedule that precedes it — the risk-by-risk account of what was bought, what has closed, and what is still live.

A project that returns an unspent allowance, with the reasoning attached, is not a project that came in under budget. It is a project that was governed properly. Those are different claims, and only one of them will stand up when someone asks.